Learning Concept Bottleneck Models from Mechanistic Explanations

• URL: http://arxiv.org/abs/2603.07343v1
• Date: Sat, 07 Mar 2026 21:09:01 GMT
• Title: Learning Concept Bottleneck Models from Mechanistic Explanations
• Abstract summary: Concept Bottleneck Models (CBMs) aim for ante-hoc interpretability by learning a bottleneck layer that predicts interpretable concepts before the decision.<n>Mechanistic CBM builds the bottleneck directly from a black-box model's own learned concepts.<n>We show that M-CBMs consistently surpass prior CBMs at matched sparsity.
• Score: 4.071515891039002
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Concept Bottleneck Models (CBMs) aim for ante-hoc interpretability by learning a bottleneck layer that predicts interpretable concepts before the decision. State-of-the-art approaches typically select which concepts to learn via human specification, open knowledge graphs, prompting an LLM, or using general CLIP concepts. However, concepts defined a-priori may not have sufficient predictive power for the task or even be learnable from the available data. As a result, these CBMs often significantly trail their black-box counterpart when controlling for information leakage. To address this, we introduce a novel CBM pipeline named Mechanistic CBM (M-CBM), which builds the bottleneck directly from a black-box model's own learned concepts. These concepts are extracted via Sparse Autoencoders (SAEs) and subsequently named and annotated on a selected subset of images using a Multimodal LLM. For fair comparison and leakage control, we also introduce the Number of Contributing Concepts (NCC), a decision-level sparsity metric that extends the recently proposed NEC metric. Across diverse datasets, we show that M-CBMs consistently surpass prior CBMs at matched sparsity, while improving concept predictions and providing concise explanations. Our code is available at https://github.com/Antonio-Dee/M-CBM.

Related papers

• Multimodal Concept Bottleneck Models [19.41992921538502]
  Concept Bottleneck Models (CBMs) enhance the interpretability of deep learning networks by aligning the features extracted from images with natural concepts.<n>We propose Multimodal Concept Bottleneck Model (MM-CBM) to address these issues and extend CBMs into CLIP.<n>Compared to existing methods, MM-CBM achieves up to 51.26% accuracy improvement on average across four standard benchmarks.
  arXiv  Detail & Related papers  (2026-06-18T07:41:50Z)
• Concepts Worth Having: Refining VLM-Guided Concept Bottleneck Models with Minimal Annotations [21.060815865649477]
  Vision-plus-Human-guided CBM (VH-CBM) is a hybrid approach that exploits both Vision-Language Models (VLMs) and a small amount of dense annotations.<n>VH-CBM predicts more accurate concepts than VLM-guided CBMs even when annotating as little as 1% of the data.
  arXiv  Detail & Related papers  (2026-05-13T10:07:11Z)
• Rethinking Concept Bottleneck Models: From Pitfalls to Solutions [53.84388497227224]
  Concept Bottleneck Models (CBMs) ground predictions in human-understandable concepts.<n>CBM-Suite is a methodological framework to address these challenges.
  arXiv  Detail & Related papers  (2026-03-05T19:37:49Z)
• Controllable Concept Bottleneck Models [55.03639763625018]
  Controllable Concept Bottleneck Models (CCBMs)<n>CCBMs support three granularities of model editing: concept-label-level, concept-level, and data-level.<n>CCBMs enjoy mathematically rigorous closed-form approximations derived from influence functions that obviate the need for retraining.
  arXiv  Detail & Related papers  (2026-01-01T19:30:06Z)
• Concept-Aware Batch Sampling Improves Language-Image Pretraining [78.53540190580189]
  Concept-Aware Batch Sampling (CABS) is a simple yet effective batch sampling framework that flexibly constructs batches on-the-fly.<n>We show that CABS significantly benefits CLIP/SigLIP model classes and yields highly performant models.<n>Overall, CABS represents a strong open-source alternative to proprietary online data curation algorithms.
  arXiv  Detail & Related papers  (2025-11-25T18:58:07Z)
• LLM Pretraining with Continuous Concepts [71.98047075145249]
  Next token prediction has been the standard training objective used in large language model pretraining.<n>We propose Continuous Concept Mixing (CoCoMix), a novel pretraining framework that combines discrete next token prediction with continuous concepts.
  arXiv  Detail & Related papers  (2025-02-12T16:00:11Z)
• Explain via Any Concept: Concept Bottleneck Model with Open Vocabulary Concepts [8.028021897214238]
  "OpenCBM" is the first CBM with concepts of open vocabularies. Our model significantly outperforms the previous state-of-the-art CBM by 9% in the classification accuracy on the benchmark dataset CUB-200-2011.
  arXiv  Detail & Related papers  (2024-08-05T06:42:00Z)
• Discover-then-Name: Task-Agnostic Concept Bottlenecks via Automated Concept Discovery [52.498055901649025]
  Concept Bottleneck Models (CBMs) have been proposed to address the 'black-box' problem of deep neural networks. We propose a novel CBM approach -- called Discover-then-Name-CBM (DN-CBM) -- that inverts the typical paradigm. Our concept extraction strategy is efficient, since it is agnostic to the downstream task, and uses concepts already known to the model.
  arXiv  Detail & Related papers  (2024-07-19T17:50:11Z)
• Sparse Concept Bottleneck Models: Gumbel Tricks in Contrastive Learning [86.15009879251386]
  We propose a novel architecture and method of explainable classification with Concept Bottleneck Models (CBM) CBMs require an additional set of concepts to leverage. We show a significant increase in accuracy using sparse hidden layers in CLIP-based bottleneck models.
  arXiv  Detail & Related papers  (2024-04-04T09:43:43Z)
• Eliminating Information Leakage in Hard Concept Bottleneck Models with Supervised, Hierarchical Concept Learning [17.982131928413096]
  Concept Bottleneck Models (CBMs) aim to deliver interpretable and interventionable predictions by bridging features and labels with human-understandable concepts. CBMs suffer from information leakage, where unintended information beyond the concepts are leaked to the subsequent label prediction. This paper proposes a new paradigm of CBMs, namely SupCBM, which achieves label predication via predicted concepts and a deliberately-designed intervention matrix.
  arXiv  Detail & Related papers  (2024-02-03T03:50:58Z)
• Post-hoc Concept Bottleneck Models [11.358495577593441]
  Concept Bottleneck Models (CBMs) map the inputs onto a set of interpretable concepts and use the concepts to make predictions. CBMs are restrictive in practice as they require concept labels in the training data to learn the bottleneck and do not leverage strong pretrained models. We show that we can turn any neural network into a PCBM without sacrificing model performance while still retaining interpretability benefits.
  arXiv  Detail & Related papers  (2022-05-31T00:29:26Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.